Carbon isotopes show snowpack acts as a valuable moisture subsidy... Cascades

advertisement
Carbon isotopes show snowpack acts as a valuable moisture subsidy to mountain forests in the Oregon
Cascades
By
Christopher J. Ratcliff
An Undergraduate Thesis Submitted to
Oregon State University
In partial fulfillment of
the requirements for the
degree of
Baccalaureate of Science in BioResource Research,
Water Resources Option,
Biosystems and Climate Modeling Option
June 10th, 2015
SNOWPACK AND FOREST MOISTURE STRESS
CARBON ISOTOPES SHOW SNOWPACK ACTS AS A VALUABLE MOISTURE SUBSIDY TO
MOUNTAIN FORESTS IN THE OREGON CASCADES
CHRISTOPHER J. RATCLIFF 1, STEVEN L. VOELKER 2, ANNE W. NOLIN 3
1. Department of BioResource Research, Oregon State University, Corvallis, OR 97330 USA
2. Department of Forest Ecosystems & Society, Oregon State University, Corvallis, OR 97330 USA
3. College of Earth, Ocean, and Atmospheric Sciences, Oregon State University, Corvallis, OR 97330 USA
Keywords: isotopes, drought, latewood, mountain forest, snowpack, climate change, water, watershed.
Primary Research Paper
1|Page
SNOWPACK AND FOREST MOISTURE STRESS
Abstract
This study examines climatological influences, particularly that of snowpack, on tree growth and
stable carbon isotope discrimination (Δ13C) from ~1980 to 2013 at two sites located in the upper
reaches of the McKenzie River watershed of the Oregon Cascade Mountains. We tested the use
of Δ13C values from latewood, corroborated by tree-ring width chronologies as precipitation
proxies to develop correlations between moisture stress and climate variables. Tree species at
each site included Douglas-fir and mountain hemlock. Interpolated meteorological and snowpack
data included snow water equivalent (SWE), precipitation, atmospheric temperature, vapor
pressure deficit (VPD), relative humidity (RH), and a metric estimating growing season length.
Significant correlations between latewood Δ13C and winter SWE at each site indicated the
importance of winter snowpack to our selected tree species (r = 0.35, r = 0.43). Late summer
precipitation and relative humidity (RH) were also significantly correlated with Δ13C (r = 0.49, r
= 0.46; r = 0.43, r = 0.44). High correlations at both sites reinforced that late summer VPD was
the primary driver of Δ13C (r = - 0.67, r = -0.61), which is often associated with moisture stress.
This was further supported by correlations between air temperature and Δ13C (r = -0.46, r = 0.47), which drives much of the variation in VPD.
Growing season length also showed
significance in mountain hemlocks at the site with longer average snowpack (r = -0.22, r = -0.44).
Moisture supplied by spring snow melt is a seasonably limited resource, nonetheless both sites
clearly showed that snowpack acts as a valuable moisture subsidy to coniferous mountain forests
in the Oregon Cascades. This study acts as a useful case study for future investigations into the
relationship between snowpack and forest health in the Pacific North West.
2|Page
SNOWPACK AND FOREST MOISTURE STRESS
Introduction
Mountain forests are crucial regulators of river and stream water discharge and carbon cycling in many
regions of the world. However, wildfires, pest outbreaks, and droughts are threats to mountain forests that
are intensifying as a consequence of global climate change (Van Mantgem et al., 2009; Williams et al.,
2013; Westerling et al., 2006). Contemporary climate models show that by midcentury the mean forestmoisture stress in the southwest United States (SWUS) will surpass that of the most severe droughts in
the past millennium (Williams et al., 2013). The present drought in the SWUS since the turn of the century
is the fifth strongest noted drought since 1000 AD (Williams et al., 2013). Similar drought patterns have
been observed over the last 6,000 years in the Pacific Northwest, via moisture-sensitive tree-ring records
and lake sediment cores (Nelson et al. 2011). The current drought in California is encroaching into the
Pacific North West potentially decreasing available precipitation and increasing water stress deficits.
Current droughts pose a significant threat to forest functionality; increasing the probability of forest fires,
pest outbreaks, and attack by tree-killing pathogens (Van Mantgem et al., 2009; Williams et al., 2013).
Recent examination of the relationship between montane forests and hydrology has established that
inter-annual variations in snowpack accumulation are positively correlated with forest productivity
(Trujillo et al., 2012; Roden & Ehleringer, 2007). Shifts toward earlier snowmelt have also been shown
to increase forest fire activity and mortality of mountain forests (Westerling et al., 2006; Trujillo et al.,
2012; Peterson & Peterson, 2001). For a 2°C increase in winter temperature, peak snow water equivalent
(SWE), the water content of snowpack, in the western Oregon Cascades is predicted to decrease by 62%,
while shifting the date of peak SWE 22 days earlier (Sproles et al., 2013; Eq. 3). Fluctuations in snowpack
associated with climate change may have a detrimental effect on mid-elevation forests of the Cascades,
environments which provide a wealth of ecosystem services including biomass based energy, renewable
resources, climate regulation, nutrient cycling, and water control and supply.
Despite research
3|Page
SNOWPACK AND FOREST MOISTURE STRESS
investigating the response of native forests to drought and climate factors, connections between decreasing
snowpack and forest productivity in the Oregon Cascades are still poorly understood.
Stable carbon isotope discrimination and tree-ring width chronologies have been used as proxies for
tree moisture stress in previous studies in the western United States (Roden & Ehleringer, 2007; Leavitt
et al., 2010). Summer precipitation was shown to influence stable carbon composition of ponderosa pine
tree-ring cellulose through the effects of plant water status (Roden & Ehleringer, 2007). The foundation
of this relationship is largely driven by stomatal closure, which constrains gas exchange between the tree
and atmosphere when trees incur drought stress (McCarrol & Loader, 2003). When stomata close due to
water stress they reduce intercellular CO2 concentrations, which in turn reduces discrimination against
13
CO2 over 12CO2 by Rubisco, a photosynthetic enzyme that carboxylates carbon dioxide (Farquhar et al.,
1982). In this way δ13C and Δ13C signals can act as precipitation proxies that will decrease or increase
according to the relative water stress of the tree. Stable isotope (δ13C) signals are expressed in permil
notation (Craig, 1957), where the 13C/12C ratio (Rsample) is held relative to a standard (PeeDee Belemnite)
that has an anomalously high 13C/12C (Rstandard) ratio as follows:
13
δ C=
Rsample - Rstandard
Rstandard
x 1000 = ‰
(Eq. 1)
Using this notation, most natural materials have a negative δ13C value. The resulting δ13C values are then
used to calculate isotope discrimination (Δ13C) as follows:
Δ13C = (δ13Ca – δ13Cp) / (1+ δ13Cp / 1000)
(Eq. 2)
4|Page
SNOWPACK AND FOREST MOISTURE STRESS
where δ13Ca is the average carbon isotope signal of the ambient atmosphere for a given year and δ13Cp is
that of plant tissue. Variation in Δ13C accounts for the long-term trend of decreasing δ13C since the early
19th century caused by fossil fuel emissions that have decreased atmospheric 13CO2 (Keeling et al., 2001;
Eq. 2). Conversion to Δ13C causes the correlations with climate variables to have an opposite sign
compared to δ13C. For example, positive correlations between δ13C and a climate element implies it is
driving moisture stress, but positive correlations with Δ13C implies that element is acting as a moisture
subsidy and inhibiting moisture stress. Latewood cellulose δ13C values have been negatively correlated
with maximum winter snowpack measurements taken from snow telemetry (SNOTEL) stations, implying
that in years with higher snowpack trees experience less moisture stress (Roden & Ehleringer, 2007).
Latewood (LW) is one of two types of distinct growth produced in tree-rings of many species. It is
formed late in the growing season, when growth rate declines and soil moisture is generally less abundant.
LW cells are smaller and have thicker cell walls than cells produced earlier in the growing season.
Earlywood (EW) or early season growth is produced during the spring and accounts for 40-80% of a rings
growth in most conifers (Domec & Gartner, 2002). EW cells are larger and have thinner cell walls than
LW. Within a tree ring, the transition from EW to LW can be abrupt or gradual, depending on the species.
In either case, the bright hues of EW provide a distinct contrast with the LW of the preceding year, leading
to the perception of concentric rings when a transverse section of a tree stem is viewed. EW cells that
developed under low water-stress conditions can better maintain higher turgor pressure (water pressure
exerted on cell walls from within) and therefore develop larger lumens and thinner cells walls. LW cells
often experience lower turgor pressure due to drier conditions, which can contribute to cells being smaller
and having thicker cell walls. These mechanisms result in the distinct differences between LW and EW.
In summer-dry ecosystems, the late summer often corresponds to the greatest moisture deficits, impacting
LW cell numbers and dimensions. This allows tree-ring width measurements, particularly LW parameters,
5|Page
SNOWPACK AND FOREST MOISTURE STRESS
to act as precipitation proxies, which in turn can support whether moisture stress, as indicated by variation
in LW Δ13C, has a significant effect on tree growth.
To our knowledge no study has directly quantified the relationship between SWE and LW isotopic
composition or tree-ring widths in the Oregon Cascades. The overarching goal of this study was to
characterize the relationship between SWE and LW Δ 13C as an indicator of late summer moisture stress
in mountain forests of the Oregon Cascades. We hypothesized that SWE of the preceding winter would
be negatively correlated with late summer moisture stress of coniferous trees.
Study Site
The study area is located in the Oregon Cascades, situated in the headwaters of the McKenzie River Basin
(MRB), a 90-mile tributary of the Willamette River and a primary watershed of the Willamette Valley
(Figure. 1). Average annual precipitation in the MRB varies greatly by elevation and ranges from 1,000
mm to 3,200 mm (Jefferson et al., 2008). Winter air temperatures remain close to 0°C, so the phase of
precipitation is very sensitive to temperature readily shifting from rain, to snow, to a mixture of both
(Sproles et al., 2013). The rain-snow transition zone is between 400 m to 1200 m (Tague and Grant, 2004).
Above 1200 m, most of the cold season precipitation falls as snow. Snowpack exceeding 2 m of SWE can
accumulate providing substantial summertime flow to the basin. The highly porous and permeable basalt
geology of the area acts as an aquifer helping maintain stream flow (Lux, 1981). This snowpack is a key
resource for the surrounding ecosystem, as well as agricultural and municipal interest. Present day
recording of the snowpack is carried out by point-based data via Natural Resource Conservation Service
(NRCS) Snowpack Telemetry (SNOTEL) (USDA, Natural Resource Conservation Service,
http://www-.wcc.nrcs.usda.gov/snow). The length of the Cascade Crest includes native tree species such
as western hemlock, Douglas-fir, and lodgepole pine. This area acts as a prime location to investigate the
6|Page
SNOWPACK AND FOREST MOISTURE STRESS
relationship between snowpack and LW Δ13C and growth in coniferous mountain forests of the Oregon
Cascades. We selected two mid-elevation SNOTEL sites (Santiam Junction and McKenzie) from which
to collect data (Figure 1). Important characteristics of each site are listed in Table 1.
7|Page
SNOWPACK AND FOREST MOISTURE STRESS
Materials and Methods
Sampling Strategy for Tree Core Selection
Tree core samples were collected in forest areas within 0.25 km of each SNOTEL site (Santiam Junction
and McKenzie). Collection sites were located on low ridges about 30 m higher in elevation than the
SNOTEL sites in order to minimize any effects due to downslope water drainage that would have reduced
water stress. At each site, we ascertained the dominant tree species and selected dominant or codominant
individuals to sample. At the Santiam Junction site, we noted that there was an old dirt road leading up to
the ridge area so we selected individual trees that were at least 10 m from this road. Tree age, width, and
crown growth were all considered when choosing trees to sample. Sampled trees were between 100 and
150 years old to avoid the ‘juvenile effect’ (Leavitt, 2010). The ‘juvenile effect’ describes the tendency of
tree rings near the pith of the tree, representing earlier years of growth, to have increased δ13C values
possibly due to soil respiration, reduced light, and proximity to the tree crown. At McKenzie (1454 m)
mountain hemlock were sampled and at Santiam Junction (1139 m) Douglas-fir were sampled. A circular
fixed plot (r = 13.8 m) was set to characterize the tree density and basal area of each stand sampled (Table
1). Tree diameter at breast height (DBH, or 1.6 m) was measured with a steel tape for each tree > 17.8 cm
DBH. The coring strategy was guided by information on carbon isotope variability gleaned from the
literature. Previous research indicates that inter-tree variability of stable carbon isotope composition
Table 1. Site Descriptions. Mean precipitation and peak SWE values for Santiam Junction and McKenzie were acquired
from SNOTEL climate records. Tree stand data was collected within 0.25 km of each SNOTEL site. Abbreviations: BA,
basal area; MAP, mean annual precipitation; MPS, mean annual peak SWE.
Site
Location
Elevation
(m)
SNOTEL
Record
(years)
Dominant
Species
BA
(m2/ha)
Density
(trees/ha)
MAP (in)
(range)
MAPS (in)
(range)
Santiam
Junction
44° 26' N
121° 57' W
1139
1979-2013
Douglasfir
95.8
635.2
74.8
(54.8 ‒ 108.4)
18.4
(2 ‒ 38.4)
McKenzie
44° 13' N
121° 52' W
1454
1982-2013
mountain
hemlock
50.4
417.9
98.2
(59.8 ‒ 152.5)
44.4
(7.3 ‒ 83.6)
8|Page
SNOWPACK AND FOREST MOISTURE STRESS
ranges from 1 to 3‰, while intra-tree (circumferential) variability ranges from 0.5 to 1.5‰ (Leavitt, 2010).
For our study, variations were minimized by sampling several cores per tree and pooling cellulose from
eight trees. For isotope sampling, we extracted two to three 12-mm cores per tree from eight trees within
each plot, at DBH (Phipps, 1985). In the vicinity of the plot, eleven additional trees were sampled with a
5.15 mm diameter increment corer to aid in cross-dating and ring-width analyses. Thus, we collected 27
cores at the Santiam Junction site and 33 cores at the McKenzie site for a total of 60 cores. Upon extraction
from the tree, each core was sealed in a plastic tube for transport back to the laboratory for later analysis.
The location of each sample site was documented using handheld GPS and each sampled tree was marked
with flagging tape in case sites needed to be revisited (Table 1).
Core Processing for Ring-Width Measurement
Cores were attached to wooden mounts using standard wood glue with the long axis of the cells oriented
vertically. Mounted cores were sanded with 180 grit paper and then 220 fine grit paper to remove scratches.
This allowed easy visualization and differentiation between EW and LW bands in each annual ring. Cores
were then dated and labeled according to age using a dot system (1 dot = 10 years, 2 dots = 50 years, 3
dots = 100 years). Manual cross-dating, matching ring width patterns between cores from the same tree
and between trees visually to ensure each ring is assigned its correct year of formation, was completed
before width measurement. Measurement of ring-width series was completed using a stereo microscope
affixed over a Velmex ring-width encoder that measures to 0.001 mm precision. The microscope was
interfaced with MeasureJ2X tree-ring measurement software on a desktop computer allowing
management of ring-width series (VoorTech Consulting, http://www.voortech.com/projectj2x/). Cores
were measured from earliest year of formation to the latest. The extent of measured years spanned from
1837 to 2013.
9|Page
SNOWPACK AND FOREST MOISTURE STRESS
Measurement and cross-dating accuracy were quality checked using COFECHA software, with a
segment size of 30 years and overlap of 15 years (Holmes, 1983; Grissino-Mayer, H.D., 2001).
COFECHA creates a master sequence from all available series to correlate with segments of individual
width series. Negative and low correlations show segments with potential errors or missing rings that need
to be reexamined. Width series were split among sites and detrended using ARSTAN (Auto Regressive
Standardization) software to produce ring-width chronologies (Cook, 1985). ARSTAN removes lowfrequency components of ring-width variation due to endogenous age-related disturbances that are not
produced by variations in climate. ARSTAN was run using batch mode processing with two curve fitting
methods applied to the width series; a negative exponential and 100 year spline curve to remove low
frequency variations (Cook and Kairiukstis, 1990).
Stable Carbon Isotope Analysis
The LW of each annual tree-ring was excised with a razor blade. Annual samples of LW were combined
so each site had pooled yearly samples for the period of record. Pooling samples was instituted to reduce
chemical preparation by 75%, while also reducing the time and costs of the study (Borella et al. 1998).
Samples were then homogenized into a uniform fine powder using a Fisher Scientific™ PowerGen™
High-Throughput Homogenizer. Over 50 mg of ground sample was loaded into polyethylene-fiber filter
bags (ANKOM, Fairport, NY), heat sealed and purified to holo-cellulose following the extraction methods
of Leavitt and Danzer (1993). Hemi-cellulose was not extracted due to small sample size left over from
pervious extractions. Year 2013 from the McKenzie site had no residual sample left. Subsequent analyses
by continuous flow isotope ratio mass-spectrometry were performed at the CEOAS Stable Isotope
Laboratory, Oregon State University to establish stable carbon isotope (δ13C) signals of ±.01 ‰ precision
(Equation 1). Cellulose samples were weighed out to 1 ± .02 mg using an ultra-microbalance and loaded
10 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
into tin capsules which were combusted in an elemental analyzer (Carlo-Erba NA1500) coupled to a
FinniganMAT delta Plus isotope ratio mass spectrometer. The resulting δ13C signals were then converted
to Δ13C signals (Equation 2).
Description of Climate Data
We acquired measured snow data from the Natural Resources Conservation Service (NRCS) SNOpack
TELemetry (SNOTEL) stations located at each of our collection sites (USDA, Natural Resource
Conservation Service, http://www-.wcc.nrcs.usda.gov/snow). These stations are part of an extensive,
automated system that measures snow water equivalent, which is a measure of the total amount of water
represented by the snowpack. Temperature and total precipitation are also measured at SNOTEL sites.
There are over 600 SNOTEL sites located in remote high-elevation watersheds of the western United
States. SNOTEL sites use a pressure sensing snow pillow to measure SWE, a storage precipitation gage
to measure snowfall plus rainfall, and an air temperature sensor. Historical measurements of mean daily
and monthly SWE and precipitation for Santiam Junction and McKenzie were obtained from SNOTEL
datasets. Peak SWE values were extracted from mean daily SWE datasets. Peak SWE, typically recorded
near the end of March or on April 1st, acts as a primary measurement to characterize inter-annual variation
in snowpack water content. Recording of daily and monthly SWE and precipitation extends back to 1979
for Santiam Junction and 1982 for McKenzie. Temperature data were more temporally limited and are
also known to have significant quality control issues so we used data from the PRISM gridded climate
data product (PRISM Climate Group Data & Products, http://www.prism.oregonstate.edu/explorer).
Datasets acquired from PRISM included monthly minimum and maximum air temperatures (Tmin and Tmax,
respectively) and dewpoint temperatures (Tdew). Average temperatures (Tavg) were calculated from the
mean of Tmin and Tmax. Water vapor pressure deficit (VPD) and relative humidity RH were derived using
11 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
temperature data and the equations listed below, where P* is saturated vapor pressure and T (Eq. 4),
represents either Tmax when used in Eq. 5, Tavg when used for Eq. 6, or Tdew.
𝑃∗ = 617.4 + (42.22 𝑥 𝑇) + (1.675 𝑥 𝑇 2 ) + (0.01408 𝑥 𝑇 3 ) + (0.0005818 𝑥 𝑇 4 )
𝑉𝑃𝐷𝑚𝑎𝑥 =
∗
∗
𝑃𝑇𝑚𝑎𝑥
−𝑃𝑇𝑑𝑒𝑤
1000
𝑃∗
= 𝑘𝑃𝑎
𝑅𝐻 = ( 𝑃𝑇𝑑𝑒𝑤
) 𝑥 100 = %
∗
(Eq. 4)
(Eq. 5)
(Eq. 6)
𝑇𝑎𝑣𝑔
Growing season length was calculated conceptually, by means of a metric estimating approach, for
each site using daily atmospheric temperature (Tavg) and SWE datasets. The number of days before July
15th since SWE was present and the number of days between January 1st and July 15th with temperatures
above 5 ºC were counted for each year. Both of these counts should represent when the growing season
begins for trees. However, even when SWE is zero, Tmin may be too low for tree growth to ensue. Data
were converted into standard scores (z-scores) as follows:
𝑧 = (𝑋 − 𝜇) / 𝜎
(Eq. 7)
where z is the z-score, X is the value of the element, μ is the population mean, and σ is the standard
deviation. Z-scores represent the number of standard deviations a datum is from the mean. Conversion of
both metrics to Z-scores allowed counts to be combined into a new Z-score simulating growing season
length. After converting to Z-scores, the degree day values were deemed to be more important so they
12 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
were multiplied by a value of 0.7 while SWE z-scores were multiplied by a value of 0.3. These weighted
Z-scores were then summed to create a value simulating growing season length.
Data Analysis
Pearson-product-moment correlation coefficients were calculated using Microsoft EXCEL to measure the
linear correlation between climate variables and ring-width or Δ13C data. SWE and other climate variables
were used as the independent variable, while LW Δ13C and detrended ring-width chronologies acted as
the dependent variables.
Spatial Dimension
To provide a broader context for our study, a spatial dimension was added to the results presenting
potential areas for future research that may experience similar trends between climate elements and treering measurements as those we observed (Figure 2). A map of the western United States was created to
show this spatial dimension using ArcGIS software (ESRI ArcGIS 10.2.2 for desktop, 2014). Map layers
depicting the extent of tree species were based off the Atlas of United States Trees, Vol. 1 (Little, 1971).
A raster of the seasonal snow zone in the western United States was acquired from the authors of Gleason
et al. 2013. The seasonal snow zone is defined by pixels with 25% or more snow cover frequency. Snow
cover frequency is defined as the ratio of the number of years when snow is observed comparative to the
number of years in the period of record (Gleason et al., 2013) Additional map layers were all collected
from the ArcGIS Online Database.
13 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
Results and Discussion
Environmental Data
Figure 3 shows mean monthly values of climate variables for both Santiam Junction and McKenzie. Trees
growing at McKenzie potentially had considerably more access to water input from winter SWE.
McKenzie had higher winter SWE and mean monthly SWE than Santiam junction during all months,
especially after the date of peak SWE, around April 1st (Figure. 3A and Table. 2). Mean (and range) peak
SWE was 18.4 (36.4), 44.4 (76.3) inches for Santiam Junction and McKenzie, respectively with more
inter-annual variation (standard deviation) at McKenzie; 9.2, 15.6 inches. Both sites experienced lower
precipitation during the months of July and August in the late summer when LW is primarily formed
(Figure 3B and Table 2). Sites experienced relatively similar mean monthly and late summer relative
humidity and maximum vapor pressure deficit with RH lowest and VPDmax highest during the late summer
(Figure 3C, D and Table 2). In general, Santiam Junction had slightly higher mean monthly and late
summer atmospheric temperatures, while also experiencing longer growing season length (Figure 3E, F
and Table 2). Temperatures were highest during the growing season at both sites (Figure 3E).
Table 2. Climate data for Santiam Junction and McKenzie from SNOTEL and PRISM datasets. Abbreviations: WSWE,
mean winter snow water equivalent; PLS, mean late summer precipitation; RHLS, mean late summer RH; VPDLS, mean
late summer maximum vapor pressure deficit; T LS, mean late summer air temperature (Tavg); GSL, mean growing season
length. Range is shown in parenthesis below mean values. Late summer (July to Sept.); winter (Dec. to March).
Site
SWEW (in)
PLS (in)
RHLS (%)
VPDLS (kPa)
TLS (°C)
GSL (z-score)
Santiam
Junction
39.2
(4.7 ‒ 85.2)
4.2
(0.8 ‒ 13.0)
55.0
(41.1 ‒ 67.6)
1.7
(1.3 – 2.2)
14.4
(12.8 – 16.1)
31.9
(21.0 – 51.6)
McKenzie
91.6
(44.4 ‒ 155.3)
5.6
(1.4 ‒ 13.3)
55.8
(44.4 ‒ 69.5)
1.9
(1.4 – 2.2)
13.1
(11.4 ‒ 15.1)
22.4
(11.5 – 37.0)
14 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
Latewood Δ13C and Width Data
Mountain hemlocks at the higher elevation site (McKenzie) had consistently less
13
C-enriched LW
cellulose than Douglas-fir at Santiam Junction resulting in higher Δ13C values (Figure 4). Mean (and
range) of LW Δ13C was 15.6 (2.3), 16.2 (1.7) ‰ for Douglas-fir and mountain hemlock, respectively. The
maximum variation in LW Δ13C between two consecutive years was 1.54, 1.06 ‰ for Douglas-fir and
mountain hemlock, respectively. This suggests that LW Δ13C represents inter-annual variation in moisture
conditions at our selected sites. Douglas-fir experienced slightly more inter-annual variation (standard
deviation) in LW Δ13C, 0.53 ‰, than mountain hemlock, 0.44 ‰. Correlation between Δ13C chronologies
at each site was r = 0.74, p < 0.00001 (n = 31), while correlation between detrended LW chronologies at
each site was insignificant; r = 0.20, p = 0.26 (n=32). High correlation between Δ13C chronologies of our
sites and species suggests our trees responded coherently to the same moisture stress signals, whereas
relation between inter-annual variations in LW width showed relatively little coherence and were unlikely
to yield a strong climatic signal. Mean (and range) of detrended latewood width chronologies was 0.98
(0.50), 0.99 (0.73) for Douglas-fir and mountain hemlock, respectively (Figure 4). Mean between-tree
variation (coefficient of variance) of detrended LW width chronologies was 0.28 for Douglas-fir and 0.37
for mountain hemlock.
Correlations between Tree-Ring Measurements and Environmental Data
Differences in correlation values between sites may be due to differences among site conditions or species
physiology. Therefore, we focus primarily on the trends observed and not differences between sites or
species. All significant correlations were either during the late summer, when LW primarily forms (late
July to September), or the winter (Figure 5). Although highly significant trends did not carry over to both
15 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
sites, positive relationships between LW width and late summer RH, coinciding with negative correlations
with late summer VPDmax and atmospheric temperature (Tavg) suggests that increasing evaporative
demand, not only increases moisture stress, but also decreases growth (Table 3). Correlations with LW
width did not corroborate observed trends shown in Figure 5 between LW Δ13C and winter SWE or late
summer precipitation. This suggests that although increased water supply influences moisture stress it
may not be sufficiently important to growth at our selected sites or to our species to show any observable
trends (Table 3). However, a number of correlations between LW width chronologies and climate
variables were indefensible, biologically and did not follow expected trends displayed in previous studies
which used similar methods (Roden & Ehleringer, 2007; Leavitt et al., 2010). As such we consider the
few unexplainable relationships with ring-width variables to be spurious, a product of using a relatively
short time series of approximately 35 and 32 years in length, reinforcing the need for a broader study
between ring growth and SWE. Originally, a multiple linear regression model incorporating LW width
analysis and stable isotope analysis was to be performed, but given the lack of fidelity among the LW
width results at the two sites, we decided to focus primarily on Δ13C as a dependent variable.
Table 3. Pearson’s correlation coefficients between climate variables and detrended latewood indices (LW) and the ratio of
LW to total ring width indices (%LW). Abbreviations: WSWE, mean winter snow water equivalent; PLS, mean late summer
precipitation; RHLS, mean late summer RH; VPDLS, mean late summer maximum vapor pressure deficit; T LS, mean late
summer air temperature; GSL, mean growing season length. Significant correlations are indicated by asterisks: * P < 0.05, **
P < 0.01. Late summer (July to Sept.); winter (Dec. to March).
Site
Parameter
SWEW
PLS
RHLS
VPDLS
TLS
GSL
Santiam Junction
LW
-0.06
0.04
0.53**
-0.22
-0.17
0.24
(n = 35 yrs.)
% LW
0.22
0.20
0.67**
-0.50**
-0.18
0.17
McKenzie
LW
0.32
0.10
0.11
-0.10
-0.15
0.02
(n = 32 yrs.)
% LW
0.28
0.32
-0.16
-0.18
-0.40*
-0.20
16 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
Figure 5 shows Pearson’s correlation coefficients between climate elements and LW carbon isotope
discrimination (Δ13C) for each site during the fall, winter, spring, and late summer. Late summer VPDmax
and air temperature (Tavg) had highly significant negative correlations with LW Δ13C implying that they
were the primary drivers of Δ13C and in theory moisture stress in our selected trees, p < 0.01 (Figure 5D,
E). For years with higher temperatures (Tavg) and VPDmax, both sites experienced more evaporative
demand, moisture stress, and lower Δ13C values. A significant correlation between winter VPDmax and
Δ13C was observed in mountain hemlock (p < 0.05). This is probably a spurious relationship since VPD
and air temperature are lowest during the winter months and should not affect tree growth or moisture
stress (Figure 3E, D). Tavg and VPDmax were positively correlated at both sites for all recorded years (p <
0.00001, Figure 3E, D). Trends between air temperature (Tavg) and LW Δ13C were similar to those of
VPDmax (Figure 5D, E). Air temperature determines if precipitation falls as rain or snow during the winter
months, influencing the relative contribution of winter precipitation that can subsidize the water budget at
each site and thereby influencing LW Δ13C during the late summer (Sproles et al., 2013). The relationship
between air temperature and VPD may have accounted for the significant correlation observed between
winter VPDmax and moisture stress in the late summer.
Growing season length showed a significant negative correlation with LW Δ13C in mountain hemlock
at the site with higher mean monthly SWE, implying that increased growing season length increases
moisture stress, p < 0.01 (Figure 5F and Figure 3A). We speculate this is because with increased growing
season length, snowpack melts earlier and trees have more time to expend available soil moisture so they
experience increased moisture stress. Mountain hemlocks at the McKenzie site, which are adapted to
snowy sub-alpine conditions, may be more sensitive to earlier snowmelt and a longer growing season
length which agrees with the findings of Peterson & Peterson (2001). Mountain hemlocks located at lower
elevation limits in the southern half of Oregon were predicted to experience increased summer drought
17 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
stress and reduced productivity as a result of earlier snowmelt and increasing temperatures from global
climate change (Peterson & Peterson, 2001).
Climate elements which inhibited moisture stress all experienced positive trends with LW Δ13C as was
expected (Figure 5). Relative humidity during the late summer showed highly significant positive
correlations with LW Δ13C in both tree species, p < 0.01 (Figure 5C). Years with cooler air temperatures
and lower VPDmax may have decreased transpiration from our selected trees during the growing period.
Late summer precipitation had a significant positive correlation with LW Δ13C at both sites implying that
late summer rain events provided valuable moisture to our selected trees, p < 0.01 (Figure 5B). Winter
SWE also showed a positive trend, supporting our hypothesis, with higher significance at McKenzie than
Santiam Junction, p < 0.01; p < 0.05 (Figure 5A). Trends with Δ13C for precipitation were the opposite of
those with SWE patterns through the fall, spring, winter, and late summer. Collectively, these patterns
suggest that in years where summer precipitation is low, trees in the Oregon Cascades will often rely on
winter SWE as a moisture subsidy during the late growing season.
Figure 6 shows trends between peak SWE and LW Δ13C. Resulting trends between peak SWE and LW
Δ13C during the period of record were not statistically significant, p > 0.05 for both sites (Figure 6A, B).
By combining z-scores of peak SWE from each site and comparing the ensuing value with a combined zscore of LW Δ13C the resulting correlation verged on significance, p > 0.05 and undoubtedly would have
reached significance with a longer period of record (Figure 5C). Thus, the sum of winter SWE provided a
more reliable indicator of snowpack’s contribution to our selected tree species than peak SWE. Figure 7
investigates this further by presenting correlations of mean monthly SWE vs. LW Δ13C overlaid by lines
which represent the accumulation and ablation of mean monthly SWE for each site. Mean December and
January SWE had the most significant correlations with LW Δ13C in the mountain hemlocks at McKenzie
(p < 0.01). Douglas-fir at Santiam Junction showed significant correlations during December (p < 0.05).
18 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
June SWE also had a significant correlation with Δ13C at McKenzie (p < 0.05), indicating that if delayed
melt events occur, they may be used by mountain hemlocks in the late summer. Sites had the highest
correlation values between SWE and LW Δ13C in December and January despite SWE peaking each year
during March or April. This trend was unexpected, but may potentially be due to large mid-winter melt
events during December and January contributing to soil moisture content before springtime melt begins
(Brooks et al., 2012). More research targeting a broader array of study sites and for a longer series of years
must be completed to further understand this unexpected observation and the relative contribution of
snowmelt to summer forest growth. Overall, our hypothesis was supported; the results from this study
show that although winter SWE melts before tree growth starts, this water source often acts as a valuable
late summer moisture subsidy to conifers in the McKenzie River Basin and montane forests of the Oregon
Cascades.
Error Estimates and Limitations
There are multiple dimensions of error to consider for this study. Inter-tree and intra-tree variation in δ13C
were considered in the sampling strategy of tree core collection so it should have caused no large variations
in Δ13C series for each site. Potential error may have developed when choosing trees to be sampled;
however, this error cannot be estimated and is negligible as many samples were collected from each site
and pooled annually into homogeneous mixtures. Samples were also taken at locations not in the exact
vicinity of the SNOTEL sites, but 0.25 km from the sites. Snow data may not be perfectly representative
of the conditions of the sites from which trees were sampled. Selected tree-core collection sites were also
chosen based on their vicinity to SNOTEL stations. In most traditional dendrochronology studies, sites
are chosen which are hypothesized to increase the sensitivity of tree growth to a primary limiting climatic
19 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
factor. In this case we studied trees nearby to the SNOTEL sites, with little regard to factors that could
constrain site water availability such as shallow soils overlaying basalt bedrock or thick volcanic scoria
or coarse pumice and ash sediments, each of which could have increased the likelihood of finding trees
with greater late-season moisture stress. Different tree species were also selected as a result of our
sampling strategy and the location of our collection sites. Differences between species in phenology, leaflevel physiology, rooting-depth characteristics, or growth allocation may have caused minor variations in
our results. However, we expect no large fluctuations in overall trends since tree species from the same
genera and located in Oregon were observed to display similar growth rates and isotopic composition
(Saffell et al, 2014). Potential errors may also have occurred during various phases of preparation and
analyses. During the preparation of LW cellulose for isotopic analysis, excision of LW by razor blade may
have been error prone and imperfect. Additionally, homogenization of powdered cellulose is never perfect.
Thirdly, the time of year represented by LW may vary somewhat from year to year (affecting the
dependent variable tree-ring Δ13C or LW proportion), while the independent variables used (snowpack
and meteorological data) have a more static timeframe. Intrinsic sample errors such as injuries or missing
rings can pose as factors for error as well; however, cross-dating by COFECHA should have eliminated
any errors during width measurement. This study was also limited to a short time frame of up to 35 years
at the most, which may have largely impacted the LW width results, which were meant to support results
from stable isotope analyses.
20 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
Conclusion
The overarching goal of this study was to distinguish the relationship between SWE and tree-ring Δ13C,
to understand how snowpack may influence late summer moisture stress of coniferous mountain forests
in the Oregon Cascades. We observed that in years with higher SWE, moisture stress was inhibited and
there was an inverse relationship between subsequent winter snowpack and late summer moisture stress,
supporting our hypothesis. Winter snowpack appears to have acted as a valuable moisture subsidy to the
mountain forests in the Oregon Cascades we investigated. We expect that there is potential extent for this
empirically-derived relationship to extend across similar elevations and in similar climatic conditions as
the Oregon Cascades throughout the western United States (Figure 2), suggesting that there are a great
many prospective locations for future research. Variations in ring width, density, structure, and the carbon,
hydrogen, and oxygen isotopic composition of tree-rings and other organic matter can be used as proxies
for a variety of climate parameters, including snowpack water content (McCarrol & Loader, 2004; Roden
& Ehleringer, 2007; Leavitt et al., 2010). These types of tree-ring measurements, and Δ13C in particular,
should therefore be seen as a valuable resource for future studies evaluating how snow hydrology affects
forest functioning.
This was the first quantitative study of how SWE might affect LW characteristics and Δ13C in the
McKenzie River Basin of the Oregon Cascades. Our results provide a valuable case study from the Oregon
Cascades that can help inform future, widespread and in-depth investigations between snowpack and
forest health. These investigations may be the most crucial in the Pacific Northwest, where snowpacks are
projected to decline drastically by mid-century due to global climate change introducing drastic shifts in
snowmelt, peak runoff, and determining whether precipitation falls as rain or snow (Sproles et al. 2013).
By mid-century the sites we used in this study, and most sub-basin watersheds below 2000 m will no
longer be capable of sustaining winter precipitation as snowpack (Sproles et al. 2013). For this reason it
21 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
is important for forest managers to consider the impending consequences of global climate change on
snowpack within the Pacific Northwest, selecting more drought resistant tree species or adapting
management practices to promote conservation of snowpack; snowpack which acts as a valuable source
of moisture to the forests they manage.
22 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
Acknowledgements
Funding for this project was provided by the Bioresource Research (BRR) Department at Oregon State
University (OSU). We thank K. Gleason for providing the seasonal snow zone raster file used in our
spatial dimension map. We thank J. McKay at the College of Earth, Ocean, and Atmospheric Sciences
(CEOAS) Stable Isotope Laboratory at OSU for her guidance when completing stable isotope analysis.
We thank the USDA Forest Service Meinzer and Woodruff laboratories for the use of their instruments
and laboratories. We thank K. Field, W. Crannell, L. Kayes, and the Mountain Hydroclimatology Research
Group at OSU for their helpful feedback and advice throughout the course of this study. Lastly, we thank
friends and family members who provided continued support.
23 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
References
1. Borella S., Leuenberger M., Saurer M. and Siegwolf R. (1998). Reducing uncertainties in δ13C
analysis of tree rings: Pooling, milling, and cellulose extraction, Journal of Geophysical Research
103: doi: 10.1029/98JD01169. issn: 0148-0227.
2. Brooks J, Barnard H.R., Coulombe R, McDonnell J.J. (2010). Ecohydrologic separation of water
between trees and streams in a Mediterranean climate, Nat Geosci 3, 100–104
3. Cook E. R. and Kairiukstis L. A. (1990). Methods of Dendrochronology: Applications in the
Environmental Sciences. Springer. ISBN-13: 978-0-7923-0586-6.
4. Cook E.R. (1985). A Time Series Approach to Tree-Ring Standardization. Unpublished PhD
Dissertation. University of Arizona, Tucson, AZ, USA
5. Craig H. (1957). Isotopic standards for carbon and oxygen and correction factors for mass
spectrometric analysis of carbon dioxide, Goechimica et Cosmochimica Acta 12, 133–149.
6. Domec J.C. & Gartner B.L. (2002a). Age- and position-related changes in hydraulic versus
mechanical dysfunction of xylem: inferrring the design criteria for Douglas-fir wood structure, Tree
Physiology 22, 91–104.
7. Farquhar G.D., O'Leary M.H., Berry J.A. (1982). On the relationship between carbon isotope
discrimination and the intercellular carbon dioxide concentration in leaves, Austral. J. Plant Physiol.
9, 121‒137.
8. Gleason K. E., Nolin A.W., Roth T.R. (2013). Charred forests increase snowmelt: Effects of burned
woody debris and incoming solar radiation on snow ablation, Geophysical Research Letters 40, 1‒8.
9. Grissino-Mayer H.D. (2001). Evaluating cross-dating accuracy: a manual and tutorial for the
computer program COFECHA, Tree-Ring Research 57, 205–221.
10. Holmes R.L. (1983). Computer assisted quality control in tree-ring dating and measurement, TreeRing Bulletin 43, 69‒78.
11. Keeling C.D., Piper S.C., Bacastow R.B., Wahlen M, Whorf T.P., Heimann M, Meijer H.A. (2001).
Exchanges of atmospheric CO2 and 13CO2 with the terrestrial biosphere and oceans from 1978 to
2000, I. Global aspects, SIO Reference Series, No. 01-06, Scripps Institution of Oceanography, San
Diego, 88 pages.
12. Leavitt S.W. (2010). Tree-ring C-H-O isotope variability and sampling, Science of the Total
Environment 408, 5244–5253.
13. Leavitt S.W., Danzer S.R. (1993). Method for batch processing small wood samples to holocellulose
for stable-carbon isotope analysis, Anal. Chem. 65, 87–89.
14. Leavitt S.W., Woodhouse C.A., Castro C.L., Wright W.E., Meko D.M., Touchan R, Griffin D,
Ciancarelli, B. (2011). The North American Monsoon in the U.S. Southwest: Potential for
investigation with tree-ring carbon isotopes, Quaternary International, v. 235, p. 101‒107.
24 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
15. Little E.L., Jr. (1971). Atlas of United States trees, volume 1, conifers and important hardwoods:
U.S. Department of Agriculture Miscellaneous Publication 1146, 9 p., 200 maps.
16. Lux D.R. (1981). Geochronology. geochemistry. and petrogenesis of basaltic rocks from the western
Cascades. Oregon: Columbus. Ohio. Ohio State University doctoral dissertation. 171 p.
17. McCarroll D., Loader N.J. (2004). Stable isotopes in tree rings, Quat. Sci. Rev. 23, 771–801.
18. Nelson D. B., Abbott M.B., Steinman B.A., Polissar P.J., Stansell N.D., Ortiz J.D., Rosenmeier
M.F., Finney B.P., Riedel J (2011). A 6,000 year lake record of drought from the Pacific Northwest,
Proc. Natl. Acad. Sci. U.S.A. 108, 3870–3875.
19. Peterson D.W., Peterson D.L. (2001). Mountain hemlock growth responds to climatic variability at
annual and decadal time scales, Ecology 82, 3330–3345.
20. Phipps R. L. (1985). Collecting, Preparing, Crossdating, and Measuring Tree Increment Cores, US
Geological Survey Water Resources Investigations Report 85-4148, 48 pp
21. Roden, J. S., Ehleringer J. R. (2007). Summer precipitation influences the stable oxygen and carbon
isotopic composition of tree-ring cellulose in Pinus ponderosa, Tree Physiology 27, 491‒501.
22. Saffell B.J., Meinzer F.C., Voelker S.L., Shaw D.C., Brooks J.R., Lachenbruch B, McKay J. (2014).
Tree-ring stable isotopes record the impact of a foliar fungal pathogen on CO2 assimilation and
growth on Douglas-fir, Plant Cell Environ 37, 1536–1547.
23. Sproles E., Nolin A.W., Rittger K, Painter T (2013). Climate change impacts on maritime mountain
snowpack in the Oregon Cascades, Hydrology and Earth System Science 17, 2581‒2597.
24. Tague C, Grant G.E. (2004). A geological framework for interpreting the low flow regimes of
Cascade streams, Willamette River Basin, Oregon, Water Resources Research, v40.
25. Trujillo E., Molotch N.P., Goulden M.L., Kelly A.E., Bales R.C. (2012). Elevation-dependent
influence of snow accumulation on forest greening, Nature Geoscience 5, 705‒709.
26. Van Mantgem P.J., Stephenson N.L., Byrne J.C., Daniels L.D., Franklin J.F., Fulé P.Z. et al. (2009).
Widespread increase of tree mortality rates in the western United States, Science 323, 521–524.
27. Westerling A. L., Hidalgo H, Cayan D.R., Swetnam T (2006). Warming and Earlier Spring Increases
Western US Forest Wildfire Activity, Science 313, 940‒943.
28. Williams A.P., Allen C.D., Macalady A.K., Griffin D, Woodhouse C.A., Meko D.M., Swetnam
T.W., Rauscher S.A., Seager R, Grissino-Mayer H.D. et al. (2013). Temperature as a potent driver of
regional forest drought stress and tree mortality, Nature Climate Change 3, 292–297.
25 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
Figure. 1 Context map of the study area created in ArcGIS displaying the McKenzie River Basin
Watershed and our collection sites in the Oregon Cascades.
26 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
Figure. 2 Map of the spatial extent of our study created in ArcGIS displaying the study area and the extent of
both tree species within the seasonal snow zone.
.
27 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
Figure 3. Mean monthly snow water equivalent, precipitation, relative humidity, maximum vapor pressure
deficit, and air temperature (Tavg) from SNOTEL and PRISM datasets for McKenzie and Santiam Junction
(Panel A – E). Panel F shows annual growing season length for each site.
28 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
Figure 4. Time series of latewood carbon isotope discrimination (Δ13C) and mean
detrended LW width chronologies for Douglas-fir (Santiam Junction) and mountain
hemlock (McKenzie).
29 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
Figure 5. Correlations between climate variables and annual carbon isotope discrimination (Δ 13C) in latewood for
Santiam Junction (dark grey, n = 35 yrs.) and McKenzie (light grey, n = 31 yrs.). SWE and precipitation are from
SNOTEL datasets, while the rest of the climate datasets are PRISM modeled. Pearson’s Correlation Coefficient
(y-axis) depicts the correlation between variables by season and between sites (x-axis). Seasons: Fall (Aug. to
Oct.), winter (Dec. to March), spring (March to May), late summer (July to Sept.). Significant correlations are
indicated by asterisks: * P < 0.05, ** P < 0.01.
30 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
Figure 6. Time series of LW Δ13C versus peak SWE for each site
(Panel A, B) and the combined z-score of LW Δ13C versus the
combined z-score of peak SWE (Panel C).
31 | P a g e
SNOWPACK AND FOREST MOISTURE STRESS
Figure 7. Pearson correlation coefficient values (bars) of mean monthly SWE vs. latewood Δ 13C for Santiam
Junction (n = 35 yrs.) and McKenzie (n = 31 yrs.). Mean monthly SWE for each site is represented by stacked
lines. Brackets are around the late growing season when latewood forms. Significant correlations are indicated
by asterisks: * P < 0.05, ** P < 0.01.
32 | P a g e
Download